OUP user menu

Amiodarone for the prevention of sudden cardiac death: a meta-analysis of randomized controlled trials

Jonathan P. Piccini, Jeffrey S. Berger, Christopher M. O’Connor
DOI: http://dx.doi.org/10.1093/eurheartj/ehp100 1245-1253 First published online: 31 March 2009


Aims Not all patients at risk for sudden cardiac death (SCD) are eligible for, or have access to implantable cardioverter defibrillator (ICD) implantation. There are conflicting data regarding the efficacy and safety of amiodarone for the prevention of SCD.

Methods and results We conducted a meta-analysis of all randomized controlled trials examining the use of amiodarone vs. placebo/control for the prevention of SCD. We identified 15 trials, which randomized 8522 patients to amiodarone or placebo/control. Amiodarone decreased the incidence of SCD [7.1 vs. 9.7%; OR 0.71 (0.61–0.84), P < 0.001] and cardiovascular death (CVD) [14.0 vs. 16.3%; OR 0.82 (0.71–0.94), P = 0.004]. There was a 1.5% absolute risk reduction in all-cause mortality which did not meet statistical significance (P = 0.093). Amiodarone therapy increased the risk of pulmonary [2.9 vs. 1.5%; OR 1.97, (1.27–3.04), P = 0.002], and thyroid [3.6 vs. 0.4%; OR 5.68, (2.94–10.98), P < 0.001] toxicity.

Conclusion Amiodarone reduces the risk of SCD by 29% and CVD by 18%, and therefore, represents a viable alternative in patients who are not eligible for or who do not have access to ICD therapy for the prevention of SCD. However, amiodarone therapy is neutral with respect to all-cause mortality and is associated with a two- and five-fold increased risk of pulmonary and thyroid toxicity.

  • Amiodarone
  • Sudden cardiac death
  • Cardiomyopathy


This risk of sudden cardiac death (SCD) is most pronounced among patients with heart failure, in whom the 1 year absolute risk of SCD is between 4 and 13%.1,2 The implantable cardioverter defibrillator (ICD) is an effective means of preventing SCD in patients with left ventricular dysfunction, however, its use is limited by cost and access to the technology.3 Not all patients at high risk of SCD are ICD candidates, including those patients with New York Heart Association (NYHA) class IV heart failure or those with a life-expectancy less than 1 year. Alternative pharmacological agents have been studied to decrease SCD. At present, there are conflicting data on the efficacy and safety of anti-arrhythmic drug therapy with amiodarone for the prevention of SCD.47 Consistent with these observations, amiodarone has a Class IIb indication (Level of evidence C) for the prevention of SCD in those patients who are not eligible for ICD implantation.8

Previous systematic reviews published more than a decade ago4,6 have shown that amiodarone is associated with decreased arrhythmic death, however, they did not incorporate more contemporary large-scale trials of amiodarone, including the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) and the Optimal Pharmacological Therapy in Implantable Cardioverter Defibrillator Patients (OPTIC) trial. Additionally, recent observational studies have questioned the safety of amiodarone in patients with left-ventricular dysfunction.9,10 Given the present guidelines, conflicting trial results, and the results of recent observational studies, there is a need to re-evaluate the cumulative evidence regarding the safety and efficacy of amiodarone, to appropriately guide clinical decisions in patients with cardiomyopathy.11


Study search

We searched MEDLINE (1966–2007), the Cochrane Controlled Trials Register, the US Food and Drug Administration website (http://www.fda.gov), and the National Institutes of Health clinicaltrials.gov database for reports of randomized controlled trials of amiodarone for the prevention of SCD. MEDLINE was searched with the following medical subject heading terms: amiodarone and randomized controlled trials. The MEDLINE query was limited to studies involving human subjects, written in English. Finally, the bibliographies of six narrative and systematic reviews were manually searched for additional citations.4,6,1215

Eligibility and data abstraction

Studies in which patients were randomized to amiodarone and placebo or inactive control were included in the final analysis. Additional inclusion criteria included: treatment for >30 days, follow-up ≥6 months, and availability of all-cause mortality as an endpoint. Studies of patients with shock-refractory ventricular arrhythmias, out-of-hospital cardiac arrest, patients <18 years, randomization to amiodarone vs. a class Ic or class III antiarrhythmic drug (without a placebo or standard of care arm) were excluded from the analysis. Finally, studies of patients with ICDs were excluded unless ICDs were present in both the treatment and control arms of the trial.

Citations were reviewed and data was abstracted independently in a standardized and unblinded fashion by two of the investigators (J.P.P. and J.S.B.). The MEDLINE query results included those reports identified by the other search methods (www.fda.gov, clinicaltrials.gov, Cochrane database, and the aforementioned bibliographies), with the exception of one trial identified in the references of published reviews.16 Abstracted data included eligibility criteria, study population demographics, baseline characteristics, study design (including the treatment and control arms), follow-up, and outcomes. Pre-specified outcomes of interest included SCD, cardiovascular death (CVD), and all-cause mortality. In the one study which included ICD patients in both the treatment and control arms, ICD discharges were included in the SCD endpoint, since they represent potentially life-threatening arrhythmic events.17 Outcomes were analysed according to intention-to-treat. Study drug discontinuation and drug toxicities, including pulmonary toxicity, thyroid toxicity, elevation of alanine aminotransferase, and aspartate aminotransferases, and symptomatic bradycardia were safety endpoints. Study quality was formally evaluated using the Delphi Consensus Criteria for quality assessment of randomized controlled trials.18 For the purpose of this analysis, studies which met six or more of the nine criteria were considered high quality. Shown in Figure 1 is the study selection process, according to the QUOROM guidelines.19

Figure 1

QUOROM flow diagram for the meta-analysis. Study selection process according to the Quality of Reporting of Meta-analyses (QUOROM) guidelines. AF, atrial fibrillation; VT, ventricular tachycardia.

Statistical analysis

The patient was chosen as the individual unit of analysis (as opposed to person years). The effects of amiodarone on SCD, CVD, all-cause mortality, and the incidences of drug toxicities were determined with both fixed-effects and random-effects modelling. Fixed-effects modelling was performed according to the Mantel and Haenszel method. Random-effects modelling was conducted according to the DerSimonian and Laird method. The results were similar with both methods, therefore, only the random-effects estimates are reported. The measure of treatment effect was reported by odds ratios (OR) with 95% confidence intervals (CI). Subgroup analyses were planned a priori, and included allocation by amiodarone dose (≥ or <200 mg/day), study population (post-myocardial infarction or heart failure), aetiology of cardiomyopathy (ischaemic or non-ischaemic), concomitant beta-blocker use (<50% and ≥50%), and follow-up duration (≥ or <12 months).We assessed heterogeneity between studies using Cochrane’s Q statistic. In order to ascertain publication bias, a funnel plot was constructed. Sensitivity analyses were performed for each outcome to assess the contribution of each study to the pooled estimate by excluding individual trials one at a time and recalculating the combined OR for the remaining studies. Statistical testing was two-tailed, and statistical significance was declared with α = 0.05. All analyses were conducted using the Comprehensive Meta-Analysis program (Biostat, Englewood, NJ).


Search results

After searching MEDLINE, the Cochrane Library, clinicaltrials.gov, and the Food and Drug Administration (www.fda.gov), we identified 308 abstracts which were reviewed for inclusion and exclusion criteria (Figure 1). Among this group of abstracts, 284 were excluded as follows: evaluation of amiodarone for the maintenance of sinus rhythm in patients with atrial fibrillation or treatment of supraventricular arrhythmias (n = 142), non-randomized study (including observational studies, pharmacodynamics/pharmacokinetic studies, substudies, editorials, etc.; n = 108), absence of a placebo or inactive control arm (n = 18), treatment for <30 days (n = 6), study involving dronedarone and other benzofuran derivatives (n = 4), shock refractory VT or out of hospital arrest (n = 3), follow-up duration <6 months (n = 2), and the inclusion of subjects <18 years (n = 1). The full manuscripts for the remaining 25 studies were retrieved for detailed review. Following full manuscript review, an additional 10 studies were excluded: no placebo or control (n = 6), duplicate report or substudy (n = 3), and follow-up <6 months (n = 1).

Trial characteristics and study quality

As shown in Table 1, we identified 15 randomized controlled trials of amiodarone for inclusion in this meta-analysis, which enrolled a total of 9716 patients.5,16,2032 Among these studies, only the patients in the amiodarone and placebo/inactive control arms were included in this analysis (n = 8522). Most trials employed two treatment arms, however, four trials employed three treatment arms.5,21,24,32 In the Basel Antiarrhythmic Study of Infarct Survival (BASIS) trial, patients randomized to individual antiarrhythmic drug therapy were excluded (n = 100/312).21 In the Spanish Study on Sudden Death (SSSD) trial patients randomized to metoprolol were excluded (n = 130/368), since beta-blockers were excluded in the amiodarone and control arms.24 Patients in the ICD arm of the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) were also excluded from this analysis (n = 829/2521).5 All patients in the Optimal Pharmacological Therapy in Implantable Cardioverter Defibrillator Patients (OPTIC) trial had an ICD, therefore the amiodarone with beta-blocker and beta-blocker alone arms were classified as the intervention and control patients, respectively. OPTIC patients randomized to sotalol were excluded from the analysis (n = 134/412).32 Eleven of the 15 (73%) trials were multi-centre and six trials enrolled patients in more than one country. The mean follow-up duration was 1.7 years (interquartile range 1.0–2.0). The majority of trials (n = 9/15) used a daily maintenance dose of 200 mg.

View this table:
Table 1

Randomized trials of amiodarone for the prevention of sudden cardiac death

Trial (Reference)YearNumber of patientsInclusion criteriaComparatorBlindDaily maintenance dose (mg)Follow-up (months)Delphi criteriaa
Hockings et al.201987200AMIPlaceboSingle-blind200126
Hamer et al.16198934NYHA IVPlaceboDouble-blind200235
Burkart et al. BASIS211990312AMI and multiform PVCs or NSVTControlOpen label200125
Nicklas et al.221991101EF < 30%, NYHA ≥ III, frequent PVCsPlacbeoDouble-blind200128
Ceremuzynski et al.231992613AMI not eligible to receive BBPlaceboDouble-blind400129
Navarro-Lopez et al. SSSD241993368AMI, EF 20–45%, frequent PVCsControlOpen label200345
Doval et al. GESICA251994516NYHA ≥ II and EF ≤ 35% or cardiomegaly, or LVEDD > 3.2 cm/m2ControlOpen label300249
Singh et al. CHF STAT271995674HF symptoms, cardiomegaly, EF ≤ 40%, frequent PVCsPlaceboDouble-blind300458
Garguichevich et al. EPAMSA261995127EF < 35%, frequent PVCsControlOpen label400126
Biswas et al.28199676NYHA ≥ III, EF < 35%, frequent PVCsPlaceboSingle-blind200126
Julian et al. EMIAT3019971486AMI, EF < 40%PlaceboDouble-blind200219
Cairns et al. CAMIAT2919971202AMI and NVST or frequent PVCsPlaceboDouble-blind400, then 300, then 200219
Elizari et al. GEMICA3120001073AMI, heart failure (Killip A or B)PlaceboDouble-blind20068
Bardy et al. SCD HeFT520052521NYHA II–III, EF ≤ 35%PlaceboDouble-blindWeight based (200, 300, or 400)45.59
Connolly et al. OPTIC322006412Resuscitated cardiac arrest or sustained VT/VF with EF < 40% or inducible VT/VF with EF < 40% or syncopeControl (BB vs. BB and amio)Open label200127
  • AMI, acute myocardial infarction; NYHA, New York Heart Association; PVC, premature ventricular contraction; NSVT, non-sustained ventricular tachycardia; EF, ejection fraction; BB, beta-blocker; LVEDD, left-ventricular end diastolic dimension; VT/VF, ventricular tachycardia/ventricular fibrillation.

  • aNumber of Delphi criteria met out of a total of nine.18

Common exclusion criteria included decompensated heart failure,5,16,20,26,31 a history of sustained ventricular arrhythmias or those requiring antiarrhythmic drug therapy,5,16,2131 and a QTc>500 ms.23,24,27,31 Using the Delphi criteria, 12 (80%) of the trials were deemed high quality with a score of 6 or greater (Table 1). Among the 15 trials, four did not report power calculations,16,20,23,28 and four failed to meet their pre-determined sample size for their primary hypothesis test.2426,31 All trials were analysed according to the intention-to-treat paradigm.

Baseline patient characteristics

Baseline patient characteristics (Table 2) were remarkably similar in age and gender in the 15 trials. The mean ages ranged from 57 to 68 and no trial enrolled more than 35% women. Seven trials were restricted to patients with ischaemic cardiomyopathy, while the remainder enrolled patients with ischaemic and non-ischaemic cardiomyopathy. All of the trials were primary prevention trials, except the OPTIC trial, which included patients who received ICDs for both primary and secondary prevention indications. Mean left-ventricular ejection fractions ranged from 18 to 44%. Beta-blocker use ranged from 0 to 100% (median 44.3%, interquartile range 24.8–63.3%) and was not reported in six trials. The Q statistic failed to indicate heterogeneity for any of the endpoints analysed.

View this table:
Table 2

Patient characteristics in randomized trials of amiodarone for the prevention of sudden cardiac death

Trial (Reference)Mean age (years)Male (%)1°/2° PreventionPopulation (Ischaemic or Non-ischaemic)Mean EF (%)Prior MI (%)Heart failure (%)Beta-blockers (%)ACE-inhibitors (%)
Hockings et al.20NR97.0PrimaryIschaemicNRNRNRNRNR
Hamer et al.1668.388.2PrimaryBoth18.066.7100NR60.5
Burkart et al. BASIS2161.083.2PrimaryIschaemic44.041.0NR24.8NR
Nicklas et al.2257.585.1PrimaryBoth20.052.5100NR63.1
Ceremuzynski et al.2359.069.7PrimaryIschaemic33.1% had EF<40%100NR0NR
Navarro-Lopez et al. SSSD2458.087.4PrimaryIschaemic35.023.917.636.68.0
Doval et al. GESICA2559.380.9PrimaryBoth19.539.0100NR90.3
Singh et al. CHF STAT2765.699.0PrimaryBoth25.371.41004.378.3
Garguichevich et al. EPAMSA2661.077.8PrimaryBoth27.040.8NRNRNR
Biswas et al.2857.068.4PrimaryBoth35.153.9100NR67.5
Julian et al. EMIAT3059.984.4PrimaryIschaemic30.210052.444.358.3
Cairns et al. CAMIAT2964.082.3PrimaryIschaemicNR10023.559.531.5
Elizari et al. GEMICA3160.477.9PrimaryIschaemicNR10010063.334.7
Bardy et al. SCD HeFT560.176.5PrimaryBoth2552.010068.771.7
Connolly et al. OPTIC3263.481.0Primary and SecondaryBoth3479.944.2100NR
  • NR, not reported; EF, ejection fraction; MI, myocardial infarction.

Efficacy of amiodarone

A total of 1607 deaths occurred among the 8522 patients included in this analysis, including 715 SCDs. The SCD rate was 7.1% (n = 302/4260) in those treated with amiodarone compared with 9.7% (n = 413/4262) in those treated with placebo/control (OR 0.72; 95% CI 0.61–0.84, P < 0.001). Cardiovascular mortality was 14.0% (n = 578 /4120) in those treated with amiodarone and 16.3% (n = 674/4124) in those assigned to placebo/control (OR 0.82; 95% CI 0.71–0.94, P = 0.004). All-cause mortality was lower in patients treated with amiodarone (18.1 vs. 19.6%), however, this difference did not reach statistical significance (OR 0.87; 95% CI 0.75–1.02, P = 0.093). Amiodarone was neutral with respect to heart failure death (OR 0.92, 95% CI 0.76–1.12, P = 0.408). Based on these data, treatment of 38 patients with amiodarone is required to prevent one SCD and 42 patients to prevent one CVD. Shown in Figure 2AC are the individual odds ratios associated with amiodarone therapy in each trial for SCD, CVD, and all-cause death. Notably, amiodarone had no significant effect on non-CVD (OR 1.17, 95% CI 0.94–1.45, P = 0.169).

Figure 2

Efficacy of amiodarone compared with control for the (A) prevention of sudden cardiac death, (B) cardiovascular death, and (C) all-cause mortality in patients with cardiomyopathy.

Despite the absence of statistical heterogeneity, since the OPTIC trial included ICD treated patients, we conducted a sensitivity analysis to assess the impact of this trial on the results. When excluding the OPTIC trial from the random-effects estimates, there was no significant difference: OR for all cause death 0.87 (95% CI 0.75–1.01), P = 0.066 and OR for SCD 0.74 (95% CI 0.63–0.87), P < 0.001. There were no CVDs in the OPTIC trial. Without SCD-HeFT, the largest trial (whose relative overall weight was 18%), the results were quite similar with an OR for all cause death 0.85 (95% CI 0.70–1.02), P = 0.081.

Safety of amiodarone

As shown in Table 3, end-organ toxicities were more frequent in the amiodarone group. Pulmonary toxicity occurred in 2.9% of amiodarone users vs. 1.5% in the placebo/control group (P = 0.002). There were six pulmonary deaths in the amiodarone group reported in the 15 trials (n = 6/8522), which corresponds to an incidence of <0.001%. Thyroid toxicity occurred in 3.6% of the amiodarone group vs. 0.4% in the placebo/control group (OR 5.68, 95% CI 2.94–10.98, P < 0.001). Additionally, hepatic toxicity (1.9 vs. 0.70%, P = 0.015) and bradyarrhythmias were more common in patients randomized to amiodarone. While cancer deaths were only reported in four trials,16,23,30,31 cancer deaths were more numerous in those patients assigned to amiodarone treatment [0.7% (n = 13/1609) vs. 0.2% (n = 4/1597), P = 0.0501].

View this table:
Table 3

Adverse events associated with discontinuation of amiodarone

Adverse reactionAmiodarone (%)Control (%)Random effects odds ratio (95% CI)P-valueNumber needed to harm (95% CI)P-value for heterogeneity
Pulmonary toxicity82/2787 (2.9)41/2777 (1.5)1.97 (1.27–3.04)0.00269 (45–144)0.383
Thyroid toxicity134/3732 (3.6)15/3724 (0.4)5.68 (2.94–10.98)<0.00132 (26–39)0.254
Hepatic toxicity35/1889 (1.85)16/2276 (0.7)2.10 (1.15–3.82)0.01587 (54–222)0.767
Brady-arrhythmia90/3245 (2.8)45/2938 (1.5)1.78 (1.16–2.72)0.00881 (51–191)0.368

Among the 4260 patients assigned to amiodarone pharmacotherapy, 1206 (28.7%) discontinued drug. When we restricted our analysis of study drug discontinuation to those nine trials which randomized against placebo (and reported placebo discontinuation), the amiodarone discontinuation rate was 31.6% (n = 1120/3545) vs. 21.1% (n = 744/3530) in the placebo group (P < 0.0001).

Treatment effects in subgroups

The treatment effect of amiodarone was analysed in several pre-specified subgroups. As shown in Table 4, there were no significant differences in the rates of SCD, CVD, or all-cause death according to amiodarone dose, indication, aetiology of cardiomyopathy, or follow-up duration.

View this table:
Table 4

Subgroup analyses for sudden cardiac death, cardiovascular death, and all-cause death

SCDCVDAll-cause death
Amiodarone dose
 ≤200 mg/day0.67 (0.48–0.94)0.90 (0.72–1.13)0.99 (0.75–1.31)
 >200 mg/day0.68 (0.61–0.90)0.77 (0.64–0.94)0.81 (0.68–0.98)
 Post-myocardial infarction0.66 (0.51–0.86)0.82 (0.67–1.00)0.88 (0.67–1.15)
 Heart failure0.72 (0.54–0.96)0.81 (0.66–1.01)0.86 (0.69–1.06)
Aetiology of cardiomyopathy
 Ischaemic0.66 (0.51–0.86)0.82 (0.67–1.00)0.88 (0.67–1.15)
Concomitant beta-blocker use
 <50%0.70 (0.54–0.90)0.77 (0.60–0.98)0.79 (0.60–1.05)
 ≥50%0.66 (0.48–0.93)0.92 (0.76–1.11)1.00 (0.81–1.23)
Follow-up duration
 ≤12 months0.61 (0.41–0.92)0.72 (0.50–1.03)0.85 (0.56–1.29)
 >12 months0.75 (0.63–0.90)0.86 (0.74–0.98)0.90 (0.79–1.02)
  • Data are shown as random effects odds ratios (95% confidence intervals).

  • SCD, sudden cardiac death; CVD, cardiovascular death.

Publication bias

In order to evaluate the impact of potential publication bias, we plotted study precision (1/standard error) against the log odds ratio for the treatment effect (prevention of SCD). As demonstrated in Figure 3, the funnel plot was largely symmetric with no evidence of publication bias.

Figure 3

Funnel plot of precision and treatment effect. Publication bias was assessed by plotting study precision (1/standard error) against the log odds ratio for the treatment effect.


In this meta-analysis of over 8500 randomized patients, there were three major findings. First, we found that amiodarone is effective for the prevention of both sudden and CVD. Second, despite statistically significant reductions in SCD and CVD, amiodarone does not appear to lead to convincing reductions in all-cause mortality. Finally, many patients are unable to continue amiodarone pharmacotherapy, a drug associated with definite risks of end-organ toxicity.

The therapeutic landscape has changed considerably since the last published systematic review of amiodarone for the prevention of SCD.4 In the interim, ICD implantation has become a Class I indicated treatment for the prevention of SCD in those with cardiomyopathy and an LVEF < 30%.8 Additionally, several large, more contemporary, randomized control trials of amiodarone have been completed.5,31,32

The history and development of amiodarone is well known to arrhythmologists. From its beginnings as an anti-anginal agent, to its subsequent discovery as an antiarrhythmic agent, the use of amiodarone has gradually increased with time. At present, amiodarone is most often used in patients with refractory atrial fibrillation and ventricular arrhythmias, especially in those with left-ventricular dysfunction. Since the advent of the ICD, the role of amiodarone in the primary prevention of SCD, has remained a less-favourable alternative, due to the absence of a clear-cut mortality benefit, and the presence of numerous end-organ toxicities, including thyroid disease, pneumonitis and fibrosis, as well as hepatic dysfunction. Recent reports have even suggested that amiodarone may increase all-cause mortality.9,10 The question remains, however, is amiodarone safe and efficacious for the prevention of SCD in patients with cardiomyopathy?

Retrospective analyses of both the Carvedilol Or Metoprolol European Trial (COMET) and the VALsartan In Acute myocardial infarction (VALIANT) Trial demonstrated that treatment with amiodarone was associated with increased mortality.9,10 Despite adjustment for heart failure severity, the results of these retrospective studies are subject to bias, specifically with amiodarone use serving as a marker of increased disease severity. In SCD-HeFT (a randomized trial included in our meta-analysis which compared amiodarone and placebo) amiodarone significantly increased all-cause mortality by 44% in patients with NYHA class III heart failure.5 Nevertheless, in this same trial ICD therapy was not associated with improved survival in NYHA class III patients, nor was the trial powered for subgroup analyses according to NYHA classification.5 In this analysis of 8522 randomized patients, amiodarone was associated with an absolute 1.5% reduction in all-cause death which did not meet statistical significance. Thus, at the very least, our results demonstrate that amiodarone does not lead to increased all-cause mortality.

The most common adverse event in patients randomized to amiodarone was thyroid toxicity. While the absolute risk of thyroid toxicity was <4%, amiodarone associated thyroid disease has been linked to significantly increased mortality, even after adjusting for LV dysfunction.33 In light of this association, it is reassuring to see no increase in non-CVD, or all-cause mortality in the overall analysis or in any of the individual trials included in this meta-analysis. Investigators have observed that amiodarone is associated with an increase in non-sudden death, and have hypothesized that this may be due to worsening LV function or conduction disturbances.34 The decrease in CVD in this report argues against these claims.

Interestingly, among the four trials which reported cancer-related death, there was increased cancer death in those treated with amiodarone. Although not adequately powered to assess differences in the development of malignancies, the discrepancy in cancer-related death merits further study.

Unfortunately, long-term adherence to amiodarone is poor. When compared with placebo, nearly 10% of the patients randomized to amiodarone discontinued therapy. Yet, despite the high discontinuation rate, our meta-analysis still demonstrated significant decreases in SCD and CVD in the intention-to-treat analyses. Reductions in SCD and CVD may be greater in those patients who successfully tolerate amiodarone. An important challenge, which remains, is the successful identification of patients who can and will be able to tolerate long-term amiodarone pharmacotherapy.


This study, as with any meta-analysis, is subject to several potential biases. First, our findings may be prone to publication bias favouring amiodarone. Although our comparison of precision and treatment effect did not identify outliers suggestive of significant publication bias,35 this technique has limited sensitivity. Secondly, our analysis was restricted to randomized controlled trials. While randomized controlled trials minimize bias and are the gold-standard for determination of experimental effect, they may not be reflective of patients treated in general clinical practice. Thirdly, our meta-analysis incorporates trials of amiodarone which span nearly two decades. During this time period, there were numerous changes in the pharmacological and non-pharmacological management of cardiomyopathy, which may limit the practical implementation of the integrated data and findings. On the other hand, the most recent systematic reviews of amiodarone are dated11 and do not reflect more contemporary trials of amiodarone. Finally, we did not have access to patient-level data and therefore relied on published results. Overall, given the large number of randomized patients in this trial and the absence of detectable publication bias, our results are likely to be robust.

Clinical implications

While amiodarone leads to considerable reductions in SCD and CVD, it is also associated with a high discontinuation rate and significant end-organ adverse reactions, including hepatic, pulmonary, and thyroid toxicity. Based upon our data, for every 1000 patients with cardiomyopathy treated with amiodarone, 15 all-cause deaths, 24 CVDs, and 26 sudden deaths would be averted. Alternatively, 32 patients would develop thyroid toxicity, 14 would develop amiodarone associated lung disease, 11 would develop hepatic toxicity, and 13 would experience bradyarrhythmias. Given this risk/benefit profile, amiodarone therapy may be reasonable in certain patients at high risk of SCD, who are not eligible for, or who do not have access to ICD implantation. These findings support the use of amiodarone advocated by the current ACC/AHA/ESC guidelines (Level IIb recommendation, level of evidence C), though our results would argue a higher level of evidence.8


Amiodarone reduces the risk of SCD by 26% and CVD by 18% in patients with cardiomyopathy but did not reduce overall mortality significantly. The high discontinuation rate and risk of end-organ toxicity remain barriers to successful utilization of this important pharmacotherapeutic. Patients with cardiomyopathy, who are at high risk for SCD but are not eligible for ICD placement, should be considered as candidates for amiodarone therapy after weighing the risk/benefit ratio.


This work was not externally funded.

Conflict of interest: J.P.P. is listed as an inventor on a US provisional patent application (021150004) covering methods for the prevention of sudden cardiac death. J.S.B. and C.M.C. declare that they have no conflict of interest.


We would like to thank Anthony Doll for pictorial assistance.


View Abstract